Required CIGS and CIGS/Mo Interface Properties for High-Efficiency Cu(In, Ga)Se 2 Based Solar Cells

In this work, we have modeled and simulated the electrical performance of CIGS thin-film solar cell using one-dimensional simulation software (SCAPS-1D). Starting from a baseline model that reproduced the experimental results, the properties of the absorber layer and the CIGS/Mo interface have been ex-plored, and the requirements for high-efficiency CIGS solar cell were proposed. Simulation results show that the band-gap, acceptor density, defect density are crucial parameters that affect the performance of the solar cell. The best conversion efficiency is obtained when the absorber band-gap is around 1.2 eV, the acceptor density at 10 16 cm −3 and the defect density less than 10 14 cm −3 . In addition, CIGS/Mo interface has been investigated. It appears that a thin MoSe 2 layer reduces recombination at this interface. An improvement of 1.5 to 2.5 mA/cm 2 in the current density (J sc ) depending on the absorber thickness is obtained.


Introduction
The thin-film solar cells represent a considerable hope in the field of PV solar cells. The goal of current research in thin-film solar cells is to develop low-cost, viable, environmentally-friendly materials which is able to compete with the conventional silicon-based structures. Among the thin-film solar cells, those

Numerical Modeling
The complexity of the solar cell design increases when efficiency enhancement concepts are considered and computer-aided design becomes necessary for novel semiconductor device development and optimization. Using powerful device simulator is an important strategy to better understand the degree of the performance enhancement that can be provided by these new device structures. Computational analysis was performed using the one-dimensional simulation program SCAPS. Given the proper device structure, values of material parameters and initial conditions, SCAPS program calculates the internal electrical parameters of CIGS solar cells by solving the system of semiconductor equation based on Poisson equation, electrons and holes continuity equations by coupled method of Newton-Rapson. Recombination currents are calculated with the Shockley-Read-Hall (SRH) model for bulk defects and an extension of the SHR model for interface defects. A deeper insight into the effect of the material properties such as doping concentration, free carrier mobilities, band-gap, and structural properties such as different layers and layers thicknesses on device performance can be obtained by the simulation. The material parameters used as the inputs are selected based on the values reported in literature [16] [17] [18]. The semiconductor properties of the intrinsic ZnO:B, i-ZnO, CdS, OVC and CIGS layers used as the input parameters for the simulations are given in Table 1.
All the layers are polycrystalline and therefore contain a large number of different defects. In our model, one type of single level defects is introduced in each layer. These are all compensating defects positioned at the intrinsic level which is close to mid-gap. Neutral interface defects for recombination were also positioned at mid-gap. Neutral cross-sections were selected in the range 10 −18 -10 −15 cm 2 , while attractive ones were selected in the range 10 −13 -10 −12 cm 2 . To pin the Fermi level at the interface level OVC/CdS, donor defects were placed 0.2 eV below the conduction band. These have small capture cross-sections to separate between pinning and the recombination parameters of the OVC layer [17]. The OVC layer parameters are similar to those of bulk CIGS except its band-gap, shallow donor density, and low carrier mobility.
The introduction of Ga into the absorber layer to form the CIGS alloy results in a widening of the band-gap from 1.02 eV to 1.67 eV which are respectively the gap of the CIS and the CGS and Equation (1) where b is the optical bowing coefficient [19]. The variation in Ga-content in the absorber also affects different material parameters such the conduction band [19], absorption coefficient [20] [21], electron affinity [22], hole mobility [23], net carrier concentration [23] [24], defect density [25]. All these parameters have been calculated according the Ga-content or taken from the literature [20]- [25].
The absorption file used in the simulation were calculated over the entire Ga-content using the equation 2πk α λ = [20], where the optical contact k as function of Ga-content is extracted over the wavelength (λ) between 300 and 1300 nm from Palson's and Alonso's papers [20] [21] and is shown in Figure   1 according to Ga-content [28], and the temperature is set at 300 K. The equivalent band diagram calculated in thermodynamic equilibrium condition is given in Figure 1(a). The J-V characteristic and quantum efficiency are represented in Figure 1(c) and Figure 1(d) respectively for x = 0.3 and compared to the experimental results [17]. This step is very important in numerical simulation, since it avoids outliers. There is a good similarity between the simulated results and the experimental results and thus validates our model.

Effect of the Absorber Band-Gap
The band-gap of the absorber layer is an important variable to enhance the per- The limitation of V oc may be due to defects that increase with the gallium rate [25]. The open circuit voltage is almost independent of the absorber thickness.
The short-circuit current density (Figure2(b)) decreases with the absorber's band-gap. This decrease is especially important when the absorber thickness is reduced. This decrease can be attributed to the reduction in the generation rate at the p-n junction due to the decrease in absorption [30] and the increase in recombination at the rear contact when the thickness is significantly reduced [7].
The efficiency (Figure 2(c)) of the solar cell as well as the fill factor (FF) ( Figure   2(d)) increase with the band-gap and the best performance is obtained for 1.2 < E g <1.3 eV. Above this value, the overall performance of the solar cell begins to drop. These results are consistent with the experimental and numerical results obtained from other simulation software [24] [31] [32]. Experimentally, the best CIGS based solar cells are obtained with a Ga-content around x = 0.3, corresponding to a E g = 1.2 eV [24]. For a gallium rate x exceeding this value, a drastic decrease in the solar cell's performance is observed. The theoretical estimation of the optimal band-gap energy to achieve the best performance in the photovoltaic devices is in the range of 1.4 to 1.5 eV for the solar spectrum of AM1.5G [30].

Effect of Acceptor Concentration
Many impurities during the growth of the absorber are likely to increase the density of acceptors within it. The most common case is the doping of the absorber with the sodium that diffuses from the soda glass substrate. The increase of the sodium doping in the absorber results in an improvement of the open-circuit voltage (V oc ) [34]. In addition, the best cells produced nowadays use an alkali post-treatment such as potassium (K) [35] [36], Cs [1]. Several material and device characterizations performed to illuminate the effects of the alkali treatment showed an increased free carrier concentration and reduced carrier recombination throughout the whole absorber film contributed to the improved performance [1]. However, it is recommended to control dopants to optimize the performance of the device since, at high levels, they could reduce mobility and consequently, the lifetime of the charge carriers. In the case of dopants from the substrate, the control is ensured by a barrier (e.g. Al 2 O 3 ) between the molybdenum and the substrate [37] [38].

Effect of Absorber Defect Concentration
Defects in the CIGS layer have a crucial role in the cell's performance. For the improvement of CIGS based solar cells device, it is important to understand the impact of the absorber quality on the cell performance and the critical range of defect density on the electrical parameters. The amount of Ga added to the alloy not only influences the band-gap energy but also the transport mechanism and the defects in the absorber [25] [39]. Many studies have shown that the quality of the absorber is the origin of the low performance of CIGS with a high Ga-content   Thereafter, we use this worst scenario to better elucidate the effect of the absorber defect density as shown in Figure 4.
The electrical parameters are less sensitive to defects when they are less than 10 14 cm −3 regardless of the thickness of the absorber. Per account, beyond 10 14 cm −3 , all parameters drop drastically. This suggests that a high defect density may be the origin of the poor performance in CIGS cells with a high Ga content. The record efficiency of 22.9% obtained by the Solar Frontier team was correlated to an improvement of the absorber quality [1].

Mo/CIGS Interface Optimization
At the Mo/CIGS interface an ohmic contact with a low contact resistance is desired in order to extract efficiently the photo-generated charge carriers from the CIGS absorber [40]. It is generally observed that a MoSe 2 layer is formed at the Mo/CIGS interface during the deposition of the CIGS absorber on the Mo by the selenization of Cu-In-Ga precursor. Several factor can be responsible for the MoSe 2 layer: the sputtering pressure of Mo [41], Na content from soda glass [42] [43] [44], the selenization temperature [45]. Many studies are unanimous that the MoSe 2 layer contributes to the improvement of adhesion at the CIGS/Mo interface [44] [46]. In addition, the MoSe 2 layer at the Mo/CIGS interface acts in a beneficial way by changing the Mo/CIGS hetero-contact from Schottky to an ohmic type contact [40] [46] [47].
Generally, the Mo/CIGS interface is a high recombination area, especially for ultra-thin absorbers. The MoSe 2 layer with its 1.4 eV band-gap could be a good electron reflector, very important to reduce Mo/CIGS interface recombination [45] [48].
However, an optimal MoSe 2 layer thickness is required since a very thick MoSe 2 limits the current collecting ability of the back electrode due to the high resistivity of MoSe 2 (10 1 -10 4 Ωcm) and hence deteriorating the electrical parameters [49] [50]. An optimization of the MoSe 2 layer thickness is therefore required to give it its role as an electron reflector. The MoSe 2 layer parameters used in the simulation are represented in Table 3. The equivalent band-diagrams with and without the MoSe 2 layer are shown in Figure 5(a) and Figure 5  This results in a good collection of charge carriers as shown in Figure 5(d).
To elucidate the advantages or disadvantages of the MoSe 2 layer, we performed the simulations by focusing only on its thickness, which seems to be the

Conclusion
We investigated the requirements to obtain high performance CIGS-based solar cells with SCAPS simulation software. Starting from a baseline model that reproduced the experimental results, we have shown that the achievement of high performance CIGS solar cell requires the optimization of the absorber's properties but also a careful focus on the Mo/CIGS interface. These requirements in the absorber process can be summarized as follows: 1) control the Ga-content to obtain a band-gap between 1.2 eV and 1.3 eV; 2) the acceptor density should be around 10 16 cm −3 and 3) the defect density must be lower than 10 14 cm −3 . The selenization conditions of Cu-In-Ga precursor on the Mo must be adjusted correctly to obtain ultra-thin MoSe 2 layer at the Mo/CIGS.